Save 7 hours of dinner planning per week (in 30 minutes)

Hello there 😎

In 2022, I worked at a healthcare tech company. We sold a behavioral app that helped people lose weight. Part of my job was running experiments to figure out what else people would buy (think: supplements, exercise equipment, digital scales, subscriptions to digital classes, etc.).

The best thing we sold was the meal plans. A few months in and these meal plans we created were doing a few million dollars per month.

But only about one in ten people who bought one cooked a single meal from it in the first week.

Call after call, the feedback was the same:

Truth was, I couldn't figure out a meal plan for my family either. We had two kids at the time. We'd moved the family to Dublin and I was working EST hours. Every night was the same: we'd finish at the creche, go to the playground, and then it'd be a scramble to get something on the table before I had to go back to work.

Why is dinner in a newsletter about work?

Three reasons:

1/ Because it is work.

Paul Millerd has a phrase for why that is easy to miss.

He says we are stuck in a job-shaped reality distortion field. His days are filled with work: childcare, cooking at home, schlepping kids, answering email.

His days are full of work, but none of it is job-shaped.

Preparing food for a family of 4 (and now 6!) is the same. Nobody pays you to plan the week, buy the food, chop the vegetables and get dinner in front of your family. Somebody still has to do it.

2/ Because it kills the two worst questions of the day.

"What are we going to eat?"

Then the one the kids start asking on a loop around 4 p.m.: "What are we doing tonight?"

Both of them stop when the answer already exists.

3/ Because when it's solved, the time goes somewhere.

It can go back to your family.

It can go to yourself.

It can go to that portfolio career you're trying to build.

Getting this figured out frees up 5-10 hours a week, and lifts the mental weight that drags at you the rest of the time.

What changes when you run a system like this

You get home at 6 p.m. with a plan. Dinner is already done. Or something's in the slow cooker for 6 hours. Or you have a takeout order you chose on purpose three days ago. There's no decision left to make in the doorway.

The groceries are already in the house. Nobody runs back out at 6:40 for a jar of sauce.

You can run this system in thirty minutes a week. Often less. My version is probably over-engineered for some (I'll give you a simpler version).

The plan flexes instead of breaking. Stuff comes up: we stay out too late, the kids have friends over, etc. You can just adjust the plan.

Then dinner ends, and there's time left over. The kids go back outside to play, or we read on the couch or watch some TV. We're not sprinting (as much!) to get everyone washed up and ready for bed.

Start here right now

I'll walk through my full stack for this below, but it's probably over-engineered for you.

As with any system you build for yourself, start simple and build on it once it's working the way you want.

Paste this into whatever chat model you already use, fill in the brackets, and paste in your calendar.

When it works, you get the next few nights planned properly with your busy evenings already ruled out, a rough sketch of the week after, and one shopping list underneath. You reply with picks.

Or install the skill and stop retyping

I turned my own version into something you can install. It's called Weekly Meal Cart, it's free, and it lives here: https://github.com/chrisminglee/remote-life-os

The first time you run it, it interviews you. Household, allergies, how many nights you'll realistically cook, which evenings are already busy, where you shop, which takeout places you order from. Then it writes your rules to a file and works from those every week after.

It runs in Claude Code as a skill. Type "help me plan dinners for next week" and it picks it up. If you use Claude on the web instead, the same file works pasted into a Project.

It will never buy anything. It builds the list, or stages the cart if you use grocery delivery, and then it stops.

How my meal planning actually runs

Mine is the same logic, wired into the services my family already uses. Here's the stack:

• Google Calendar for what the week looks like

• Home Appétit, a Philly meal delivery service, for Monday through Wednesday dinners

• Roam Research for our recipes, about thirty mains, all tagged and searchable

• Instacart for groceries, Aldi first and ACME as the backup

• AnyList for the running household list, which we add to by shouting at Alexa

• Claude Code running a custom command on a schedule, which is the part that ties it all together

Thursday morning. It reads my Google Calendar about twelve days out. Anything blocking the 4 to 8 p.m. window gets flagged first. So if we have rock climbing or soccer practice that night, it plans around it.

Then it pulls this week's Home Appétit menu, ranks the entrées against what we've ordered before, and picks recipes from Roam we haven't cooked in the last few weeks. It plans the next four days properly and sketches the week after. Only those four days go on the grocery list, so nothing sits in the fridge for ten days. It emails me the whole thing with everything numbered.

Thursday evening. I reply with picks, like "1, 3, 4, 7." It fills the Home Appétit cart and the Instacart cart, writes the recipe ingredients to AnyList so the whole house can see what's coming, and stops at the checkout button. I click pay, and we're basically done.

Friday morning, if I never replied, it stages a default order so the cutoff doesn't pass empty.

Saturday and Sunday morning, it sends me one reminder: start the slow cooker by nine, or order from this place by six.

The system works about 90% of the time; the other 10% is where I come in with my feedback.

My kids are picky eaters. When they decide they hate something, I tell the system and it stops proposing it.

On weekends, the system breaks a bit differently. If I plan to cook on a Saturday it usually falls apart, because by the evening we'd rather be doing something as a family than standing at the stove. So I adjust. Weekends get a slow cooker meal now, started in the morning and ready whenever we get back.

Having the plan written down is what let me see the crack in the first place. Then I fill it in, and the next version is just a bit better.

I spent a year selling people better plans.

Every one of those customers already had the menu, the grocery list, the calorie count, and the macros. They still couldn't get dinner on the table.

Having a plan and executing a plan are two different problems, and I'd only ever sold them a solution to the first one.

There wasn't a great, accessible solution to the second problem. Now that anyone can build and share AI workflows, there is.

Get the blueprint. The system, the starter prompt and the installable skill are free at https://github.com/chrisminglee/remote-life-os

🤝 When you’re ready, here’s how I help you land your next job:

I love meeting new people. Say hi on LinkedIn 👋